# Alternatives to Google Agent Development Kit (ADK) > Source: https://theagentsindex.com/google-adk/alternatives (current datasheets, evidence-ranked) Candidates share the baseline category or are previously researched peers. Requirements, preserved strengths, explicit trade-offs and research coverage decide the order; paid tier does not. ## Shortlist ### 1. AgentsKit.js TypeScript foundation for AI agents: runtime, tools, memory, RAG, model adapters and headless chat UI bindings. **Research coverage:** 34/35 criteria **Where it improves on Google Agent Development Kit (ADK):** - [Writes code: yes — A bundled coder persona implements features from written specs using scoped filesystem and allowlisted shell tools, aimed at real multi-file TypeScript rather than snippets.](https://www.agentskit.io/docs/agents/skills/coder) - [Reviews code: yes — Two review personas ship in the skills package and pair with the GitHub tool, returning approve, request-changes or comment verdicts on a diff someone else wrote.](https://www.agentskit.io/docs/agents/skills/code-reviewer) - [In your pipeline: yes — Eval results render as JUnit, Markdown or GitHub annotations that can fail a pull request, and a GitHub Actions snippet shows the wiring.](https://www.agentskit.io/docs/production/evals/ci) **Trade-offs against Google Agent Development Kit (ADK):** - [Ships it: partial — There is no deploy feature; a GitHub Actions integration can trigger workflow_dispatch and the allowlisted shell can run git, so shipping is a pipeline you wire yourself.](https://www.agentskit.io/docs/agents/tools/integrations/github-actions) - [Data residency: partial — Jurisdiction is yours to choose because nothing is hosted by the vendor, and the router can pin selection to a region, but no residency guarantee is published.](https://www.agentskit.io/docs/api/adapters/interfaces/RouterOptions) [View AgentsKit.js](https://theagentsindex.com/agentskit) ### 2. LangChain Open source agent frameworks plus LangSmith, the platform for tracing, evaluating and deploying agents. **Research coverage:** 34/35 criteria **Where it improves on Google Agent Development Kit (ADK):** - [Writes code: yes — dcode, the vendor's terminal coding agent, writes and edits files from a natural-language task, showing diffs first.](https://docs.langchain.com/oss/deepagents/code/quickstart) - [Done-for-you service: partial — You run the software; Enterprise adds LangChain engineers, architectural guidance and team trainings, not delivery of the agent.](https://www.langchain.com/pricing) - [In your pipeline: yes — dcode runs headless with turn and time budgets and exit code 124 for CI; Engine opens pull requests on a connected GitHub repo.](https://docs.langchain.com/oss/deepagents/code/quickstart) **Trade-offs against Google Agent Development Kit (ADK):** - [Ships it: partial — Engine stops at a pull request for human review; agents themselves ship through LangSmith Deployment's 1-Click Deploy, checked for Plus and Enterprise.](https://docs.langchain.com/langsmith/engine-github) - [Self-hosted: partial — Platform self-hosting and hybrid are Enterprise-only; Developer and Plus are Cloud, while the open source packages run wherever you install them.](https://www.langchain.com/pricing) - [Getting out: partial — Bulk trace export to S3 in Parquet is Enterprise-only for accounts opened after 3 August 2026; SDK trace queries stay open to all, and credit expiry is not stated.](https://docs.langchain.com/langsmith/data-export) [View LangChain](https://theagentsindex.com/langchain) ### 3. Agno Python SDK, open-source AgentOS runtime and hosted control plane for building and operating agents. **Research coverage:** 35/35 criteria **Where it improves on Google Agent Development Kit (ADK):** - [Writes code: yes — Ships a CodingTools toolkit whose documented job is editing files and running commands in a scoped project directory.](https://docs.agno.com/use-cases/coding-agents) - [Debugs and fixes: yes — Bug fixing is a documented pattern: reproduce, patch, rerun the failing tests.](https://docs.agno.com/use-cases/coding-agents) - [Reviews code: yes — Code review over a diff is one of the documented coding-agent patterns.](https://docs.agno.com/use-cases/coding-agents) **Trade-offs against Google Agent Development Kit (ADK):** - [Ships it: partial — Publishing is deliberately outside the default loop; commits, PRs and deploys must be exposed as separately gated tools.](https://docs.agno.com/use-cases/coding-agents) - [Agent permissions: partial — Runs are unattended unless a tool is marked for approval; then the run pauses in your database until an admin resolves it.](https://docs.agno.com/agent-os/approvals/overview) - [In your pipeline: no — The enumerated serving surfaces are REST, MCP in Claude/ChatGPT and chat apps; no pull-request or CI surface appears.](https://www.agno.com/) [View Agno](https://theagentsindex.com/agno) ### 4. Letta Stateful agent harness, cloud and SDK for agents with persistent, git-versioned memory. **Research coverage:** 35/35 criteria **Where it improves on Google Agent Development Kit (ADK):** - [Writes code: yes — Agents edit and create files in your repository, including from GitHub issues and PRs.](https://docs.letta.com/platform/github-action/index.md) - [Debugs and fixes: yes — Vendor states coding agents debug failures and run test suites.](https://docs.letta.com/handbook/index.md) - [Reviews code: yes — Reviews diffs on demand or on every PR via the GitHub action's prompt input.](https://docs.letta.com/platform/github-action/index.md) **Trade-offs against Google Agent Development Kit (ADK):** - [Ships it: partial — Commits, pushes and opens pull requests; it cannot approve PRs and no deployment step is claimed.](https://docs.letta.com/platform/github-action/index.md) - [MCP server: no — MCP runs inbound only: Letta sessions consume stdio, HTTP and SSE MCP servers; no Letta MCP server is published.](https://docs.letta.com/agent-sdk/mcp/index.md) - [Audit log: no — No administrative audit log is published; the Enterprise list stops at RBAC, SSO, quotas and dedicated support.](https://docs.letta.com/pricing/index.md) [View Letta](https://theagentsindex.com/letta) ### 5. Mastra Open-source TypeScript framework for building AI agents, with a hosted platform for running and observing them. **Research coverage:** 35/35 criteria **Where it improves on Google Agent Development Kit (ADK):** - [Done-for-you service: partial — You run the software, but Mastra sells professional services (Concierge, forward-deployed engineering) and enterprise support covering planning, prototyping and workshops.](https://mastra.ai/terms-of-service) - [In a browser: yes — Studio is a hosted web UI for running agents, editing prompts, annotating traces and building datasets; it runs locally at port 4111 or deployed for the team.](https://mastra.ai/ai-agent-framework) - [How it meters: yes — Plan fee plus consumption metering: observability events, CPU hours, egress, memory tokens, retrieval storage and database rows all have per-unit overage rates. Not per seat.](https://mastra.ai/pricing) **Trade-offs against Google Agent Development Kit (ADK):** - [Data residency: partial — Self-hosted Enterprise keeps traces, prompts and outputs inside your own VPC; for the hosted platform the vendor publishes no region or jurisdiction.](https://mastra.ai/pricing) - [Getting out: partial — You own your data and can cancel any time, but retention is capped by plan (15 days to 6 months), backups are your responsibility and downgrading may delete data over the new limit.](https://mastra.ai/terms-of-service) [View Mastra](https://theagentsindex.com/mastra) ### 6. Pydantic AI Typed Python agent framework: any model, any surface, with a harness for long-running coding and research agents. **Research coverage:** 34/35 criteria **Where it improves on Google Agent Development Kit (ADK):** - [Writes code: yes — The first-party Harness package ships Coder, a combined capability giving an agent workspace-rooted read/write/edit/search file tools plus an allowlisted shell.](https://pydantic.dev/docs/ai/harness/coder/) - [Debugs and fixes: yes — The harness is written for exactly this: an agent with files, shell and planning turned loose to fix a codebase over a long session.](https://pydantic.dev/docs/ai/harness/) - [In a browser: yes — agent.to_web() and clai web serve a browser chat for any agent; the same page scopes it to local development and points production frontends at the UI event-stream protocols.](https://pydantic.dev/docs/ai/guides/web/) **Trade-offs against Google Agent Development Kit (ADK):** - [Ships it: partial — It stops at the working tree: it runs whatever commands you allowlist in the workspace, and no deploy, release or pull-request step is documented.](https://pydantic.dev/docs/ai/harness/coder/) - [Agent permissions: partial — You choose the ceiling: shell tools take allow/deny lists and default only to a destructive-command denylist, and individual tools can be marked as requiring human approval.](https://pydantic.dev/docs/ai/harness/shell/) - [In your pipeline: no — The surface table enumerates code, terminal, web chat, your frontend, editors and voice; nothing acts on a pull request or in CI, and the roadmap line names messaging channels and API endpoints as what is still missing.](https://pydantic.dev/docs/ai/overview/interfaces/) [View Pydantic AI](https://theagentsindex.com/pydantic-ai) ## Method Unknown data is never treated as a benefit or a trade-off. Featured placement is labelled but cannot alter recommendation order.